What we found — and what each finding teaches
This page is a collection of things we found while examining the published evidence — the trials, the reviews, the regulatory records and the market around them — and what those findings teach about evaluating medical information. A finding about one therapy is often a check that works on any claim.
Thirty-three of the things we found are below: twenty-eight about what happened, and five about why any of this is being studied at all. Some of them concern the therapies this site’s own routes are about. Nearly all are already published here, inside the route or the page they belong to; what is new is that they are together, and that each one says what it actually means.
You are not expected to read all of it. It is built to be searched, not read in order: find the finding you came for, take it, and go. Every finding below has its own link.
What this page is not.
It is not a ranking of therapies. The order is by the kind of question each finding answers, not by how bad anything is.
It is not a list of reasons not to pursue treatment. Several of these findings cut the other way, and the ones that do are stated as plainly as the rest.
It is not a substitute for the clinical routes. Those are where the evidence for a specific condition lives, with a physician’s signature on it — cancer · knee and joint pain · immune function · wound healing. This page is a collection of things we found while examining the evidence, and what those findings teach about evaluating medical information.
How each finding is organised. Three parts, in the same order every time:
- What happened — the fact, and where it comes from.
- What it means — including what it does not mean. A negative finding can be overread as easily as a positive one, and the middle part exists to stop both.
- What you can do with it — the check that transfers, so the same finding is useful the next time you meet a claim we have never seen. In the last section, where the findings are about biology rather than about events, the third part changes verb: it is what it lets you spot.
Where these come from. A finding is eligible only if it already appears on a page that has been through review — the clinical routes, the therapies library, or the regulatory page — or if it has been verified again against its primary source and reviewed before being added here. Every finding also has to be followable: two candidates were dropped at the last pass because their sources could not be traced to anything a reader could open, and one of those two was a finding in this field’s favour. Our internal evidence register is raw material, not vetted evidence, and interesting is not the same as verified.
No company is named anywhere on this page. The facts below are described precisely enough to be looked up; the sources carry the names. We are describing a market, not prosecuting anyone in it.
A measuring stick, before you start.
One autologous cell therapy has been through the complete regulatory path in the United States. Two multicentre randomised double-blind vehicle-controlled trials in 421 people, a defined indication — the appearance of moderate to severe nasolabial fold wrinkles in adults — and a licence granted by the FDA in 2011. At six months the physician-assessed response was 33% against 7% in one trial and 19% against 7% in the other.
That is what the finished path looks like: one specific product, one named condition, controlled trials, and a measured effect that is real and modest. None of the therapies discussed on this page has done this.
It is here as a ruler, not as a credential. The indication is part of the ruler — a cell therapy that completed the path completed it for the appearance of facial wrinkles, and knowing that is what stops the word “approved” from doing work it has not earned.
Source: FDA licence BLA 125348; trial results in Smith SR et al., 2012, PMID 22409385.
What an authorisation says, and what it does not
Approved in Europe in 2018, withdrawn in December 2024#
What happened. A stem cell therapy for a complication of Crohn’s disease was authorised in the European Union in 2018, on the strength of a trial in which it beat placebo. A larger confirmatory trial was then run. In it, the placebo group did just as well. The marketing authorisation was voluntarily withdrawn in December 2024, and the European regulator concluded that clinical benefit was no longer demonstrated.
What it means. This is what an honest evidence system looks like from the inside — a first result that did not survive a second, larger test. It does not mean the original approval was dishonest, or that the patients treated in between were harmed. It means one positive trial was not enough to settle it, which is the reason confirmatory trials exist at all. A therapy can look real, get approved, and then turn out not to be.
What you can do with it. When you are shown a positive trial, ask whether it has been confirmed by a larger one — and, if it has, what the larger one found.
Source: European regulator and manufacturer statement, 13 December 2024. Already on this site in Immune function and Why Mexico.
The first FDA approval in this category came nine years after Japan’s#
What happened. Japan approved a mesenchymal stromal cell product for graft-versus-host disease in 2015. Canada approved one. The FDA approved its first in December 2024.
What it means. The therapy did not change in December 2024. The paperwork did. Approval is applied for, not assigned: someone has to run the trials, assemble the file and pay for the process, and a product with no company behind it may never be submitted regardless of whether it works. That does not make an unapproved product the equal of an approved one — it means that “not approved here”, on its own, tells you about a filing rather than about evidence.
What you can do with it. When you hear “not FDA-approved”, ask which of four things it means: nobody has submitted it, it is approved somewhere else, it is under investigation and genuinely uncertain, or it was tested and did not hold up. They are not equivalent.
Source: FDA approval notice, 18 December 2024; peer-reviewed literature for the 2015 approval. Already on this site in Why Mexico.
No NK cell product is approved as a medicine by any major agency#
What happened. As verified across four agencies in 2026 — the American, European, Japanese and Korean regulators — no NK cell product is approved as a medicine anywhere. This is the therapy our cancer route is about.
What it means. “Not approved” is not “shown not to work”: it means that to date no authority has granted approval for that indication. It is not meaningless either — it means the question has not been settled by the system that settles it. Both statements are true at the same time.
What you can do with it. Ask for the indication and the country. “Approved”, without naming the condition it was approved for, is an incomplete sentence.
Source: approval registers of four agencies, consulted 9 August 2026; it can change. Already on this site in NK cell therapy for cancer and Why Mexico.
An approval belongs to one molecule, not to a word#
What happened. Thymalfasin — a defined, synthetic, 28-amino-acid peptide — is an approved prescription medicine in more than 30 countries for hepatitis B and C, with a safety record across more than 11,000 subjects in over 30 trials. Three chemically distinct things are routinely called by the same name: thymalfasin, a crude thymic extract, and anything labelled a “thymic peptide”.
What it means. The approval is real, and it shows the underlying idea is not absurd: immune signalling genuinely can be modulated by a defined molecule. What it does not do is transfer to a different vial with a similar name. The trap here is identity, not efficacy: it is what allows an undefined “thymic” product to be offered on the strength of an approval that is not about it.
What you can do with it. Ask whether the published evidence is for this specific preparation or for the category, and insist that the answer name a molecule rather than an adjective.
Source: review of more than 11,000 subjects across more than 30 trials (Dinetz et al., 2024). Already on this site in Therapies you may encounter and Why Mexico.
What a single trial can support
The strongest published NK result: 37 people, and no comparison group#
What happened. In the clearest published trial of engineered NK cells in lymphoid cancers, 37 people were treated. About 49 of every 100 responded, and about 68 of every 100 were alive at one year. About 32 of every 100 were free of progression at one year, which means most progressed.
What it means. The same trial carries the encouraging number and the limiting one, and both belong to it. A response is not a cure, and short responses are common. Without a comparison group there is no way to know what would have happened otherwise — which is not a reason to dismiss the result, since early trials are how anything begins, but is a reason not to read it as a rate you can expect. These are also averages across a group of people: they cannot predict what happens to one patient.
What you can do with it. Ask for the denominator and whether there was a control arm. Then ask which number stopped being quoted after the first year.
Source: Nature Medicine 2024 (Marin et al.), PMC10957466. Already on this site in NK cell therapy for cancer.
A safety pattern that has held across many small trials#
What happened. In that same trial, nobody developed cytokine release syndrome, neurotoxicity, or graft-versus-host disease — the serious complications that limit some other cell therapies. That pattern has held across many small NK trials.
What it means. A trial does not need a comparison group to document adverse events observed in the patients who received the therapy, although a small trial cannot establish the frequency of uncommon harms or the therapy’s overall safety. This is the same trial as the finding above — weak evidence of efficacy, and useful evidence about the safety experience observed in that trial. What it does not establish is that the therapy is safe in general, or that a product prepared some other way behaves like this one.
What you can do with it. Ask what a trial was able to answer, question by question. A study can be a poor source for one and a good source for another, and “small” is not a verdict on all of it.
Source: Nature Medicine 2024 (Marin et al.), PMC10957466. Already on this site in NK cell therapy for cancer.
The cells did exactly what they were designed to do, and the disease did not change#
What happened. In a trial in children with acute myeloid leukemia, the cells expanded as intended in 17 of 21 children — and there was no reduction in relapse and no improvement in survival.
What it means. Expansion is not benefit. A mechanism doing what it was meant to do is not the same as the disease behaving differently, and a trial can succeed biologically and fail clinically at the same time. It does not prove the approach never works: one negative trial, in one disease, in 21 children, is exactly that and no more.
What you can do with it. Ask whether a reported result is something that happens to the patient, or something that happens on a laboratory report. Both are worth measuring; only one of them is the reason you would take a treatment.
Source: J Immunother Cancer 2019;7:81 (Nguyen R et al.), PMID 30894213 — 21 children against a comparison group of 55. Already on this site in NK cell therapy for cancer.
A phase 3 trial with a placebo group does exist#
What happened. In knee osteoarthritis, a randomised, double-blind, multicentre, placebo-controlled phase 3 trial enrolled 146 patients, 73 per arm, and gave a single injection of allogeneic bone-marrow mesenchymal cells or placebo into the joint. Pain, stiffness and function scores were significantly better than placebo at six and twelve months. Five adverse events were judged possibly or probably related, all of them injection-site swelling and pain that resolved within days.
What it means. Most cell therapies have never reached a phase 3 trial with a placebo group, and this one did. That is a real result and a narrow one: it belongs to one product from one manufacturer, and it does not transfer to a different preparation with the same description. It is also a single trial — the thing the findings above say not to over-read.
What you can do with it. When a clinic cites “phase 3 evidence”, ask which product was in the syringe in that trial, and whether it is the one in front of you.
Source: Am J Sports Med 2023;51:2254-66 (Gupta et al.). Already on this site in Knee and joint pain.
The largest recent wound trial found nothing — in the hardest wounds#
What happened. Two hundred and twenty patients with complex diabetic foot ulcers, including exposed bone or tendon and controlled bone infection. Healing was 66 of every 100 with amniotic membrane against 60 of every 100 without — not a meaningful difference, at 26 weeks or at 50.
What it means. This is not the opposite of the positive trials. It is a boundary. The positive trials enrolled simpler ulcers, so the most useful reading is that membrane appears to help relatively uncomplicated wounds and has not been shown to help the most difficult ones. A null result in the hardest cases does not erase the benefit seen in easier ones, and the benefit seen in easier ones does not extend to the hardest.
What you can do with it. Before applying a trial’s result to yourself, read who was enrolled in it. The population is part of the finding, not background to it.
Source: Wound Repair Regen, December 2025 (Caporusso et al.), doi 10.1111/wrr.70110. Already on this site in Wound healing.
A survival difference large enough that the trial was stopped early#
What happened. In a randomised phase 2 trial in 34 people with stage 4 metastatic pancreatic cancer, median overall survival was 16 months with chemotherapy plus high-dose intravenous vitamin C, against 8 months with chemotherapy alone. The trial was stopped early for benefit.
What it means. This is one of the largest differences on this page, and its caveat belongs inside the finding rather than after it: trials stopped early for benefit systematically overestimate the effect, because they stop at the moment the difference looks largest. The trial was also small and open-label. None of that makes the result false — it makes it a reason to run the larger trial, which is what an early stop is for.
What you can do with it. When you are told a trial was stopped early because the treatment was working, treat the size of the reported benefit cautiously, and ask what the confirmatory trial found.
Source: Redox Biology 2024;77:103375 (Bodeker KL et al.), NCI-funded. Already on this site in Therapies you may encounter.
A trial that stopped with its safety question unresolved#
What happened. A small intravenous mesenchymal-cell trial in critically ill patients with ARDS was terminated after all six enrolled participants died.
What it means. The trial does not establish that the cells caused the deaths: these patients were critically ill, there was no demonstrated causal attribution, and the sponsor reported that it could not determine whether the intervention represented a safety risk. What the record shows is that the safety question remained unresolved when the study stopped.
What you can do with it. When a trial has stopped, ask what it established before it did. A termination records a question left open; on its own it answers nothing in either direction — and a registry entry will tell you which of the two you are looking at.
Source: public clinical-trial registry record, NCT04390152 (trial start 2020, listed completion 2022), verified against the registry on 18 August 2026. This is the one finding here that was not already published elsewhere on this site; its interpretation above is the reviewing physician’s.
What a summary of many trials can support
A 2026 analysis whose literature search closed in August 2024#
What happened. A 2026 analysis of 14 randomised trials in 1,056 patients found amniotic membrane roughly 1.8 times more likely to produce complete healing than standard care alone. Its literature search closed in August 2024 — so it does not include either of the two large trials published in 2025.
What it means. This does not make the analysis wrong, or its authors careless. Every review has to close its search somewhere, and the gap between the search and the publication is normal. It means a review is a photograph of a date, and the date on the cover is not that date.
What you can do with it. Check the search cutoff rather than the publication year, then ask what has been published since. It is usually stated in the methods, in one line.
Source: pooled analysis of 14 randomised trials, 2026. Already on this site in Wound healing.
Direction and certainty are two different answers#
What happened. Across randomised trials, amniotic membrane heals more wounds than standard care alone. An independent Cochrane review of 17 trials found a similar direction with a smaller effect, and rated the certainty of that evidence low.
What it means. Both are true, and they answer different questions: one is about direction, the other about certainty. A low certainty rating is not a negative result and not an accusation — it describes how much the estimate could still move as better trials arrive. Treating the two as one number is how a real effect gets oversold and how a weak one gets dismissed.
What you can do with it. Ask for both answers separately. “Does it work?” and “how sure are we?” have different evidence behind them.
Source: pooled analysis and independent Cochrane review. Already on this site in Wound healing.
A pooled result that held its size across six randomised trials#
What happened. A 2025 meta-analysis of randomised controlled trials of mesenchymal cells in knee osteoarthritis found a standardised improvement in pain, stiffness and function of −1.35 at twelve months, with a confidence interval running from −1.97 to −0.74 — a moderate to large effect that does not cross zero.
What it means. An interval that stays on one side of zero means the direction was consistent across the trials pooled, which is a genuine finding and not a formality. It is still a class result: it describes mesenchymal cell injections as a family, not any one manufacturer’s preparation. And a large standardised effect on a questionnaire is not the same as a large difference you would notice on the stairs. As discussed below, not all of the improvement measured in these trials can necessarily be attributed to the cells themselves.
What you can do with it. Look at whether a pooled interval crosses zero before you look at the headline number — and ask what the questionnaire measured, in units you recognise.
Source: J Orthop Surg Res 2025, doi 10.1186/s13018-025-06190-4 — six randomised trials, 300 patients. Already on this site in Knee and joint pain.
A pooled benefit that disappeared when only the good studies were counted#
What happened. A 2025 meta-analysis of a thymic peptide found a pooled benefit that disappeared when the analysis was restricted to high-quality studies. The largest phase 3 trial was negative: 28-day mortality of 23.4% against 24.1%.
What it means. The restriction did not prove that no effect exists. It showed that the pooled benefit was no longer apparent when the analysis was limited to the studies judged to be at lower risk of bias.
What you can do with it. Ask what happens to the effect when only the larger, higher-quality, multi-centre trials are counted. If nobody has asked that question, that is the question.
Source: 2025 meta-analysis, PMID 40969554, and the phase 3 trial. Already on this site in Therapies you may encounter and Immune function.
An effect that survived exactly that test#
What happened. Intravenous alpha lipoic acid, 600 mg daily for three weeks, was pooled across four randomised, double-masked, placebo-controlled trials in 1,258 people with symptomatic diabetic polyneuropathy. The improvement in symptoms and in neurological deficits was significant and clinically meaningful, and a separate appraisal graded the recommendation A.
What it means. This is the same test the thymic peptide failed, applied to a different compound, with the opposite outcome: restricted to double-masked placebo-controlled trials, the effect was still there. That is what a result surviving its own quality restriction looks like — and it is specific: a defined dose, a defined route and one condition, not a general claim about infusions.
What you can do with it. When a pooled result survives restriction to the better trials, that is worth more than a larger effect that does not. Ask which of the two you are being shown.
Source: meta-analysis of four trials, PMID 14984445; grading in PMC3272801. Already on this site in Therapies you may encounter.
A measured share of the improvement was not the treatment#
What happened. Across randomised trials of mesenchymal cells for knee osteoarthritis, people who received them reported meaningfully better pain, stiffness and function scores at twelve months. A 2025 analysis then measured how much of that improvement is attributable to contextual effects — the injection itself, the attention, the expectation of getting better. The share is substantial.
What it means. It is real improvement that patients genuinely feel; it just is not evidence about the cells. It does not make the therapy fake and it does not make it proven. Some people find “part of this is the ritual” disqualifying; others reasonably say that relief is relief. Both readings are defensible, and the measurement is what lets you have the argument at all.
What you can do with it. Ask how much of the expected improvement is contextual — and whether the trial you are being shown had a placebo arm in the first place.
Source: GRADE-appraised meta-analysis of placebo-controlled randomised trials, 2025, doi 10.3389/fmed.2025.1636181. Already on this site in Knee and joint pain.
The strongest human evidence for exosomes is in skin, and it was registered in advance#
What happened. A 2025 systematic review of human studies in dermatology, registered on a public review registry before it was carried out, found scar thickness reduced by 32.5% against 19.9% in controls, and elasticity improving 11.3% where controls worsened. Two of its secondary outcomes — hydration and pigmentation — were not statistically significant.
What it means. A review registered before it is run is harder to steer toward a conclusion, which is why the registration belongs in the finding rather than in a footnote to it. What the review covers is skin: it says nothing about wound healing, where there are no completed human trials at all, and nothing about any other use. The two non-significant outcomes are quoted here for the same reason the review reports them.
What you can do with it. Check whether a review was registered before it was carried out — and check what it was about before you let it stand behind a different claim.
Source: systematic review, doi 10.3390/reports8040268, PMID 41441544; registration PROSPERO CRD420251108123. Already on this site in Therapies you may encounter.
What is not in the published record
15 of 17 trials had industry involvement — and so did the ones that found nothing#
What happened. In the Cochrane review of amniotic and placental membrane, 15 of 17 trials had industry involvement, and the pattern of results suggested that small negative trials were not being published. The trials that found nothing were industry-funded too.
What it means. Industry funding is common in this field and does not, by itself, invalidate the findings. It is, however, one reason independent confirmation is important. And the second half is the part that gets skipped: the funding did not only push results in one direction, because the negative trials were funded as well. Knowing who paid tells you to look for confirmation. It does not tell you which way the error runs.
What you can do with it. Ask who funded it — and then ask a different question: has any independent group reproduced it? The second question is not a substitute for the first, and the first is not an answer on its own.
Source: independent Cochrane review of 17 trials. Already on this site in Wound healing.
Trials that were run, and never reported#
What happened. An independent Cochrane review of stem cell injections for knee osteoarthritis rated the certainty of the evidence low to very low, and found up to three larger randomised trials that were conducted and withdrawn before reporting. A pivotal randomised trial of one device system was registered and has published no results. Radiographic progression was not assessed in any included study.
What it means. That is what suspected publication bias looks like from the outside. It does not prove the missing trials were negative — nobody can know that, and that is exactly the problem. What it means is that the published record is not the whole record, and that the average of what was published is likely to be kinder than the average of what was run.
What you can do with it. Look the therapy up in a public trial registry and check whether the registered trials reported their results. A registration with no results is information.
Source: Cochrane Database Syst Rev 2025;4:CD013342.pub2 (Whittle SL et al.), PMID 40169165. Already on this site in Therapies you may encounter.
Where the documented harm has come from
The infections were traced to manufacturing, not to the concept#
What happened. In 2019 the FDA issued a public safety notification after patients in Nebraska developed serious infections from an unapproved exosome product — fewer than five people, within a two-month window. The state Attorney General later sued the clinic. Warning letters issued since have repeatedly cited failure to validate sterility.
What it means. The harm was traced to how the product was made, not to the underlying idea — and across this field, manufacturing is where the documented harm has come from more often than the biology. These are also findings about the products and manufacturers named in them. They do not describe every product of that type, and reading them as if they did would be the same mistake this page argues against, pointed the other way.
What you can do with it. Ask where it was made and under what licence, and what testing each batch gets before it is released. A cell product made in a licensed facility with documented release testing is a different object from a vial produced in a back room, even when the label says the same word.
Source: FDA public safety notification, 6 December 2019. The pattern of warning letters since sits in the agency’s enforcement record, and we have not been able to pin it to one citable reference — we would rather say so than drop the sentence or imply a source we cannot hand you. Already on this site in Therapies you may encounter and Why Mexico.
What a set of case reports establishes, and what it does not#
What happened. Delayed-onset granulomas in seven to eight women aged 26 to 52 after intradermal injection of exosome products for cosmetic purposes. Four further women with persistent redness, nodules and granulomatous inflammation, all with incomplete resolution and residual scarring. Skin necrosis after injection of a lyophilised formulation.
What it means. These describe what happened to specific patients, not what typically happens. They establish that it can happen; they do not establish how often, and on their own they do not establish that the product caused it. The absence of a frequency is not the same as a low frequency — and it is not the same as a high one either.
What you can do with it. When you are given a risk, ask where the number comes from: a trial with a denominator, or reports without one. Then ask the same question of any reassurance you are given.
Sources: Dermatological Reviews 2024 (Nahm et al.), doi 10.1002/der2.242; J Cosmet Dermatol 2025 (Park et al.), doi 10.1111/jocd.70520; J Cosmet Dermatol 2024 (Tawanwongsri et al.), doi 10.1111/jocd.16206. Already on this site in Therapies you may encounter.
An injection is a different object from a supplement#
What happened. An injectable NAD product was voluntarily recalled in July 2025 after endotoxin contamination, and in October 2025 the FDA classified that recall as Class 1 — the category reserved for products with a reasonable probability of serious harm or death. Intravenous B-complex carries documented warnings of its own: urticaria, breathing difficulty, wheezing, angioedema, and rare anaphylaxis.
What it means. The pattern here is not “vitamins are dangerous”. It is that the risk belongs to the route and to the manufacturing, not to the molecule. The same compound taken by mouth is a different object, with a different risk.
What you can do with it. When the same compound exists in oral form, ask what the injection adds — and, separately, who made the one in front of you.
Source: FDA Form 483, 18 July 2025; recall initiated 30 July 2025; FDA Class 1 classification, 21 October 2025. Already on this site in Therapies you may encounter.
In the trials themselves, the adverse events were comparable#
What happened. Across the randomised trials of amniotic and placental membrane in chronic wounds, the adverse events reported were comparable between the group receiving the membrane and the group receiving standard care alone.
What it means. Trial safety findings are real and they are narrow. They describe what happened to people enrolled in studies, with products made for studies, under protocols written for studies. Clinical trial safety should not be interpreted as a guarantee that every product or provider in the marketplace is equally safe. Both halves matter — a treatment can have a clean trial record and a market full of products that never went through one.
What you can do with it. Ask whether a safety claim comes from a trial or from a marketplace. Different populations, different products, different question.
Source: adverse events as reported in the randomised trials pooled for this route. Already on this site in Wound healing.
A safety record at a scale trials never reach#
What happened. Intravenous vitamin C is given to roughly 10,000 people a year with minimal adverse events reported; lethargy, fatigue, nausea and vomiting occur in fewer than 1%.
What it means. Volume of use tells you something a trial cannot, because rare harms only appear in large numbers of people. It tells you nothing about whether the treatment works — that is a separate question with separate evidence, and on this page that evidence is one small trial that was stopped early. Wide use with few reported harms is a genuine finding and a limited one: reported is not the same as occurred.
What you can do with it. Keep “is it safe?” and “does it work?” apart, and notice when a seller answers the second question with evidence for the first.
Source: safety record across approximately 10,000 people treated annually. Already on this site in Therapies you may encounter.
What happened in the market, which is a different question
Of 76 clinics audited, one matched the official registry on both fields#
What happened. A published audit of 76 stem cell clinics in one Mexican border city found that 13 claimed to be licensed. Six matched the official registry by name or by address. One matched on both.
What it means. The audit describes the clinics it examined. It is not a statement about every clinic operating in Mexico, and reading it as one would be the same mistake this page argues against, pointed the other way. What it is good for is the method it exposes. Seven of the thirteen matched the registry on neither field. Of the six that did match, only one matched on both — so a licensing claim can also survive a check on one field and fail on the other, which is the failure a quick check misses.
What you can do with it. Verify a clinic by name and by address, against the regulator’s own registry — not against what the website says. Checking one field is what most people do, and it is what this audit shows is not enough.
Source: Chavez J, Shah NA, Ruoss S, Cuomo RE, Ward SR, Mackey TK. Online marketing practices of regenerative medicine clinics in US-Mexico border region: a web surveillance study. Stem Cell Res Ther 2021;12:189, PMID 33736697. Already on this site in Why Mexico.
Spending went from 256 million dollars to over 10 billion while patients roughly doubled#
What happened. Medicare spending on these products in the United States went from 256 million dollars in 2019 to over 10 billion in 2024, while the number of patients treated roughly doubled. In January 2026 the payment system was rebuilt to cut that by close to 90 percent.
What it means. None of that means the products do not work. The trials are real and some of them are positive. It means the volume of use in this field ran far ahead of the evidence for it, and that a patient being offered one of these products is entering a market where that happened. How much something is used is not evidence of how well it works, in either direction.
What you can do with it. Keep the two questions apart: “how widely is this used?” and “what has it been shown to do?” have different answers, from different sources, and the first is often offered as if it answered the second.
Source: HHS Office of Inspector General report on Medicare Part B payment trends for skin substitutes, October 2025; and the CMS physician fee schedule final rule for 2026 (CMS-1832-F, 31 October 2025), which states the change is expected to reduce spending on these products by nearly 90 percent. Already on this site in Wound healing.
Fifteen and a half years, and fourteen, for grafts nobody needed#
What happened. In October 2025 two people were sentenced to fifteen and a half years and to fourteen years in prison for a 1.2 billion dollar scheme applying unnecessary grafts to elderly and hospice patients — in some cases to people who died within days. Separately, the company that funded several of the positive trials cited on our wound page had its former chief executive convicted of securities fraud.
What it means. This is misconduct in a market, not evidence about a therapy. Putting the two together would suggest that fraud in the selling of a product and the safety findings of a trial are evidence of the same thing, and they are not. Fraud tells you nothing about whether a product works. It tells you a great deal about who you may be buying it from, and those are two separate things to find out.
What you can do with it. Ask both questions, and do not let either answer stand in for the other: what has this been shown to do, and who is the person offering it to me?
Source: United States Department of Justice — sentences handed down on 7 October 2025 (fifteen and a half years) and 10 October 2025 (fourteen years). Already on this site in Wound healing.
Why any of this is being studied at all
Everything above is about things that happened. These five are about biology: the reasons serious researchers work on these approaches despite the uncertainty. They are not evidence that anything works. A plausible mechanism is the weakest rung on the ladder of evidence, and it stays weak however good the story is. They are here because a mechanism you understand is the fastest way to catch a claim that contradicts it — so the third part of each one changes verb: not what you can do, but what it lets you spot.
NK cells act on a target they have never seen#
What happened. Most immune cells have to be shown a target first, learn it, and then act. NK cells read the surface of a cell for signs that something is wrong — signals a healthy cell displays and a damaged or infected one loses — and respond to their absence. They also do not need to be matched to the patient the way T cells do, and in the trials so far they have not caused graft-versus-host disease.
What it means. That combination is why the field is interested: cells that need no matching, and that have not caused the complication which makes donor T cells dangerous, can be prepared from a healthy donor in advance rather than made for one person. It is a reason to run trials. It is not a result — the trials above are what the results look like.
What it lets you spot. That “ready from a healthy donor” is a property of the biology, not a logistical achievement. A provider charging extra for personalised matching in an NK product is selling something the biology does not ask for.
Already on this site in NK cell therapy for cancer.
The membrane is unusually invisible to the immune system#
What happened. Amniotic membrane spends nine months holding a barrier intact between two organisms while carrying signals that tell tissue to grow and tell inflammation to settle. It is naturally low in the markers the immune system uses to recognise foreign tissue, which is why it can be used between unrelated people, and it arrives with that signalling already in it rather than having it added.
What it means. That is why researchers thought a stalled wound might restart under it: the material is not a synthetic dressing with something applied on top. It explains the interest and predicts nothing about the outcome — the trials on this page point both ways, and the biology was just as true in the trial that found nothing.
What it lets you spot. That the interesting property belongs to the tissue itself. If the explanation you are given is about a proprietary processing method rather than about the tissue, ask what the method adds, and where that was measured.
Already on this site in Wound healing.
In a joint, the hypothesis is that the cells change the environment — not that they rebuild it#
What happened. Mesenchymal cells can become bone or cartilage in a laboratory, which is where the word “regenerative” comes from. That is not what they are believed to do inside a joint. Osteoarthritis is also a low-grade inflammatory process involving the lining, the fluid and the bone underneath; injected cells release signalling molecules that damp that inflammation, and most of the cells are cleared within weeks.
What it means. The hypothesis being tested is that they change the environment of the joint, not that they rebuild it. That is a much narrower claim than the word “regenerative” suggests — and it is the claim the trials were built around, which is why they measure pain and function rather than a rebuilt joint.
What it lets you spot. Anyone promising to regrow your cartilage is describing something the mechanism does not claim and the trials did not measure. Most of the injected cells are gone within weeks; whatever the injection does, it is not building a new joint surface.
Already on this site in Knee and joint pain.
The idea behind exosomes is the signal without the cell#
What happened. Much of what a cell therapy appears to do may come not from the cell taking up residence, but from what it secretes. Exosomes are the vesicles that carry that secreted signal. If the signal is what matters, it could in principle be produced, characterised and dosed like a product — no living cells to keep alive, no donor matching. Sixty-six clinical trials of these vesicles were registered between 2014 and 2024.
What it means. That is a real biological question, and it is why the field is large. It is also a conditional: the entire appeal depends on the preparation being characterisable, and across the field these preparations are not standardised, so dose and content vary between products sold under the same name.
What it lets you spot. That the premise of the approach is a specification. If the person offering it cannot tell you what is in the vial and at what concentration, they are not delivering the idea — they are invoking it.
Already on this site in Therapies you may encounter.
At high intravenous doses, vitamin C stops behaving like an antioxidant#
What happened. At the doses reached intravenously — far above anything achievable by mouth — vitamin C acts as a pro-oxidant, generating hydrogen peroxide in tissue. Laboratory work suggests cancer cells may be less able to clear it than normal cells.
What it means. That is the biological question the trials are testing, and it is a mechanism rather than a result: it explains why the intravenous and oral forms are studied as different things, and it shows clinical benefit for neither. It also cuts against a claim in the other direction — a warning about antioxidants during radiotherapy does not transfer automatically to a compound that is not acting as an antioxidant at these doses.
What it lets you spot. That the oral form is not a cheaper version of the same thing. Anyone selling a supplement as equivalent to the infusion is describing a different mechanism, whether or not they know it.
Already on this site in Therapies you may encounter.
Where to go from here
This page does not tell you what to do. It has no advice in it about any particular condition, because a finding about a literature is not a finding about a patient.
If you are trying to decide something, the clinical routes are where the evidence for a specific condition lives, with what is known, what is not, and who it is unlikely to help: Cancer · Knee & Joint Pain · Immune Support · Wound Healing.
If what you want is the state of the evidence for a therapy that has no route here, that is Therapies you may encounter. If the question is regulatory — what approval is, and what “not approved in the U.S.” can mean — that is Why this is available in Mexico.
And if you are deciding whether to explore anything at all, How to decide is genuinely a page about deciding not to, as often as deciding to.
If you are citing this page, every finding above names its source and links to the page where it already appears with its full context — which is where the number means what it says.